Are arch width measurements from Invisalign arch width tables reliable?

Despite being widely used for treatment planning, the accuracy of Invisalign arch width tables has not been independently assessed. The objective of this study was to assess whether the predicted and observed arch width changes calculated from Invisalign tables are consistent with measurements obtained from STL models.

Thirty-five adults treated with Invisalign aligners were retrospectively selected. Arch width at the maxillary and mandibular canines, premolars, and first molars was measured on digital models (pretreatment, prediction and first-refinement) using Geomagic Control X. Predicted and observed expansions (difference between predicted or post-treatment and pretreatment arch widths), and their discrepancy, were compared with the corresponding values calculated from the ClinCheck arch width tables. Three references were selected: (1) the projection of the long axis of the tooth on the occlusal surface, (2) the buccal/mesiobuccal cusps, and (3) the most lingual point of the gingival margin. Normality was assessed with the Shapiro-Wilk test. Agreement was evaluated using Bland-Altman analysis with mixed-effects models to account for clustering of repeated measurements.

840 arch widths were analyzed (35 patients, 4 tooth pairs, 2 jaws, and 3 time points). Non-normality of inter-method differences was observed in predicted expansion (gingival and occlusal references) and in discrepancy (cusp reference) (p < 0.05). Non-parametric Bland-Altman analysis showed high agreement between Geomagic and ClinCheck measurements for predicted expansion, observed expansion, and discrepancy, with bias values ranging from - 0.49 to 0.2 mm.

ClinCheck arch width tables showed strong agreement with independent metrological assessment, particularly with occlusal reference points [bias: 0.00 mm; limits of agreement: -0.80 to 1.01 mm].

ClinCheck arch width tables have been validated with an independent metrological assessment (Geomagic Control X). Predicted expansion, observed expansion, and discrepancy derived from arch width tables agreed with independent measurements using virtual casts.